Download leukemia - Emerson Statistics Home

Documentation for Clinical Trial Idarubicin in AML
Page 1 of 7
(Note: The following description is taken from Emerson and Banks in Case Studies in
Biometry (Lange, et al., eds, copyright 1994 by Wiley).)
Acute Myelogenous Leukemia
Leukemias are cancers in which blood-forming cells undergo changes resulting in
uncontrolled, malignant growth. Patients with leukemia exhibit excessive numbers of
abnormal white blood cells in their circulation. Normally, white blood cells play a major
role in the body's defense against infection. The white blood cells that predominate in
leukemias, however, are generally quite immature and limited in their ability to fight
infections. Furthermore, the cancerous cells replace the normal blood-forming cells in the
bone marrow, and thus a patient with leukemia will often present with anemia, low
numbers of platelets in the circulating blood, and other hematologic abnormalities.
The causes of leukemias have not been fully identified, though there is some evidence to
suggest that at least some cases of leukemia may be the result of exposure to
environmental factors such as ionizing radiation, chemicals such as benzene, or certain
viral infections. There is also some evidence that genetic factors may play a role.
There are several types of leukemia. They are classified by the type of blood-forming cell
which has become malignant (lymphocytic versus myelogenous), as well as the
aggressiveness of the disease (acute versus chronic). Medical treatment of leukemia
varies with the exact classification of the disease. Thus, treatment of acute myelogenous
leukemia differs from the treatment of other leukemias.
The main treatment for acute myelogenous leukemia is administration of anti-cancer
drugs which target rapidly dividing cells. Because these drugs are not perfectly selective
for cancer cells, there are generally high levels of toxicity associated with cancer
chemotherapy. In the process of killing the cancer cells, some non-diseased cells are also
affected, especially those with a naturally high rate of growth, e.g., in hair follicles, the
blood-forming cells in the bone marrow, and the cells in the lining of the gastrointestinal
tract. The doses of anti-cancer drugs which can be administered are limited by the extent
of this toxicity. Thus, leukemia chemotherapy usually proceeds through several phases.
In the first phase, termed the remission induction phase, an intensive course of
chemotherapy is administered in hopes of eliminating the leukemic clones. If this
treatment is successful, microscopic examination of the circulating blood and bone
marrow cells will reveal few or no cancer cells, and the patient is then said to have had a
remission induced. In some cases, multiple courses of remission induction chemotherapy
must be administered before a remission is achieved. At times, the remission induction
treatment fails, either because no remission is achieved after many such induction
therapies, or because the patient dies during early courses of chemotherapy.
A short period after the patient is judged to be in remission, further courses of intensive
chemotherapy are often administered in an effort to kill any remaining leukemic cells.
These additional courses are termed consolidation treatments. Following consolidation,
patients generally continue to receive periodic, less intense courses of chemotherapy
Documentation for Clinical Trial Idarubicin in AML
Page 2 of 7
called maintenance therapy. These maintenance treatments can continue for years after
initial diagnosis, and often include treatments which target areas of the body that are
known to be poorly treated during the induction phase, e.g., the brain, spinal cord, and
testes.
Despite the consolidation and maintenance therapies, leukemia often reappears in patients
who have had a successful induction of remission. In such cases of leukemic relapse, the
patient will often undergo additional remission induction treatments, sometimes with
different drugs. With each succeeding relapse, however, the success rate of remission
induction decreases. In recent years, bone marrow transplantation has been used to treat
patients who have had one or more relapses. There has also been some investigation of
the use of bone marrow transplantation prior to leukemic relapse while the patient is still
in remission. In a bone marrow transplant, all blood-forming cells in the patient are
destroyed either by chemotherapy, radiation, or both. Non-diseased bone marrow is then
transplanted into the patient in the hopes that the transplanted marrow will be able to
provide the blood-forming cells needed to ensure adequate response to infections.
While much progress has been made in the treatment of some leukemias (most notably
childhood acute lymphocytic leukemia), the success rates for the treatment of adult acute
myelogenous leukemia are not as impressive. Much research into experimental
treatments continues in the hope of finding the optimal treatment regimen.
Clinical Trials
The general strategies just described for the treatment of acute leukemias were identified
through a long series of experiments. Ideally, each aspect of the treatment regimen is
tested for its effectiveness before it is adopted for use in the general population. In this
way, the possibility is minimized that anecdotal evidence might cause ineffective, or even
harmful, treatments to be adopted. For instance, a typical scenario for the determination
of which anti-cancer drugs should be used in the remission induction phase might include
the following steps. First the drug is tested in the laboratory on cells grown in culture.
Promising drugs are then screened in animals. Those few drugs that still look effective
are then evaluated in clinical trials involving human volunteers. The highly toxic nature
of the drugs demands that such testing proceed cautiously.
There is much literature about the proper procedures by which clinical trials should be
conducted (see, for instance, Pocock, 1983). Briefly, the clinical trials of new anti-cancer
drugs proceed through three phases. In Phase I testing, the new treatment is given to
small numbers of human volunteers in order to find appropriate doses. Often Phase I
trials are conducted in patients for whom standard treatments have failed. Phase II trials
investigate preliminary measures of treatment efficacy along with further assessment of
treatment safety in slightly larger numbers of patients. Traditionally, Phase II trials do not
attempt to compare the new treatment to an existing treatment, instead they focus on
demonstrating any anti-cancer activity of the drug.
Documentation for Clinical Trial Idarubicin in AML
Page 3 of 7
Treatments which show promising anti-cancer activity at tolerable doses are then tested
in a larger Phase III trial. Such trials usually have as their goal the comparison of the new
treatment to some existing standard treatment or, if no standard therapy exists, to a
placebo. Whenever possible, these trials are conducted by randomly assigning patients to
receive either the new or comparative treatment in a double blind fashion. The double
blinding, wherein neither the patient nor the researchers directly assessing treatment
outcome know which treatment the patient is receiving, is an effort to avoid patient or
researcher bias affecting the outcome of the experiment.
Each patient randomized to receive either the new or existing treatment is followed for
treatment outcome. Typically, there are multiple ways to measure treatment success, e.g.,
length of patient survival, percent of patients surviving for two years, or percent of
patients experiencing tumor regression. Important secondary measures of treatment
success might include percent of patients experiencing toxicities and the severity of those
toxicities. At the conclusion of the study, these and other endpoints are analyzed, and the
results are used to establish the net benefit of the new treatment.
A Clinical Trial of Idarubicin in Acute Myelogenous Leukemia
In 1984, a comparative clinical trial was initiated to test the difference in therapeutic
effect between two chemotherapies indicated for the treatment of adult acute
myelogenous leukemia. Patients from a single institution, the Memorial Sloan Kettering
Cancer Center, were prospectively randomized with equal assignment probability to two
treatment arms. Since it was not possible to make the two treatments identical in
appearance, the investigators and the patients were not blinded as to which treatment the
patients received. However, the pathology data used to determine remission status was
reviewed without knowledge of treatment assignments.
The two treatments being compared were each of the class of chemicals called
anthracyclines. One of the treated groups was given the newly synthesized anthracycline
idarubicin, while the other treatment group (or treatment arm) received the standard
anthracycline agent, daunorubicin. As is common in many cancer chemotherapy
regimens, the usual induction treatment of acute myelogenous leukemia involves multiple
drugs. Thus, both treatment arms were dosed with cytosine arabinoside (ARA-C) in
addition to the anthracyclines.
The trial permitted both induction and consolidation phases of treatment. Induction
therapy consisted of an intravenous bolus of 25 mg/M2 of ARA-C, followed immediately
by 200 mg/M2 of continuously infused ARA-C for five days. The anthracyclines were
given by slow intravenous injection for the first three days of the ARA-C dosing period.
The idarubicin and daunorubicin doses were 12 mg/M2 and 50 mg/M2, respectively.
Patients failing to achieve complete remission status following this initial induction
course received a second induction course identical in schedule to the first. Patients not
responding to the second course were considered treatment failures and were
subsequently withdrawn from the study.
Documentation for Clinical Trial Idarubicin in AML
Page 4 of 7
Complete responders (i.e., patients achieving complete remission) proceeded to the
consolidation phase of the study following a 3 to 4 week rest period. The treatment plan
for the consolidation phase was the same as that used for the induction period, except that
the ARA-C was given for four days with the anthracycline administration coinciding with
the first two days of ARA-C infusion. A maximum of 4 consolidation courses could be
given to each patient.
The study protocol was amended early during the trial to omit a maintenance treatment
period and to modify criteria for identifying patients to receive bone marrow
transplantation. The maintenance course, offered to only a few patients, consisted of
randomizing patients who remained in complete remission one month after their first
consolidation course to receive either no treatment or a regimen of subcutaneous ARA-C.
Originally, patients 40 years of age or less with a suitable donor were eligible for bone
marrow transplantation after achieving a complete response. No consolidation therapy
was to be given. Under the revised design, all potentially eligible patients in complete
remission received one course of consolidation therapy. Then patients 50 years of age or
less proceeded to bone marrow transplantation if a suitable donor was available, or, if
not, were eligible for bone marrow harvest and autologous bone marrow transplantation,
that is, re-infusion of their own bone marrow cells. A patient declining autologous
transplantation, could undergo bone marrow harvest with the opportunity for autologous
bone marrow transplantation should relapse occur and a second complete remission be
achieved.
As indicated by the study protocol described above, great care was exercised in the
conduct of the clinical trial to promote comparability of the treatment arms. The primary
study objective was to demonstrate a difference between the treatment arms with respect
to the rates of patients achieving a complete remission. Secondary objectives included the
assessment of differences between the treatment arms with respect to patient toxicity
rates and patient survival. It is not uncommon that the intuitively more interesting
endpoint of patient survival be relegated to secondary status due to the larger sample size
and longer study duration required to achieve adequate statistical power to detect
clinically important differences.
During the early portion of the trial, the data were reviewed for patient safety and
compliance. The data were also casually monitored for treatment efficacy. Periodic
monitoring of the data in a clinical trial is usually indicated to satisfy the ethical
considerations involved in human experimentation. It is important that a trial be
terminated if there is reliable evidence that patients are being harmed by continuing the
study.
In the case of this study, however, the protocol, though well specified in most other
respects, initially failed to adequately specify the methods whereby the data would be
monitored. If decisions to terminate a clinical trial are based on repeated analyses of the
accruing data, the type I statistical error of hypothesis tests may be inflated over the
Documentation for Clinical Trial Idarubicin in AML
Page 5 of 7
desired level, confidence intervals may not have the correct coverage probabilities, and
bias may be introduced into the estimates of treatment effects. Though the researchers
retroactively imposed a formal stopping rule on the study, the FDA was reluctant to trust
that the final reported study results were not unduly affected by the early unplanned
analyses which showed a favorable trend for the idarubicin arm.
Sampling Scheme
Patients were accrued to the study and randomly assigned to one of the treatment arms.
The initial study protocol called for a single formal analysis of the data following the
accrual of the entire planned sample size. However, after early, unplanned analyses of the
data suggested some advantage of idarubicin over daunorubicin in inducing complete
remission, a two-sided level 0.05 O'Brien-Fleming (1979) group sequential design with a
maximum of 4 analyses was adopted at the time that 69 patients had been accrued.
Though only three formal analyses were planned for the study, the O'Brien-Fleming
design based on 4 was chosen to account for the unplanned early looks at the data that
had already taken place.
Additional analyses were then planned after groups of approximately 40 response
evaluable patients had been randomized and assessed for their remission status. The
maximum sample size was adjusted upward so that the resultant test would still have 80%
power to detect absolute differences in complete response rates of 0.20. The new estimate
called for 160 response evaluable patients. A slight excess of patients was to be enrolled
to ensure that sufficient numbers of evaluable patients were accrued, bringing the
maximum sample size to be accrued to 90 patients per treatment arm.
The first formal analysis was performed after a total of 90 patients had been accrued, and
the second (and last) such interim analysis was performed after 130 patients had been
accrued in June, 1989. At the second analysis, 10 of the patients were judged nonevaluable for determination of complete response (primarily due to improper diagnosis at
study entry), however, regulatory agencies most often want reports on the entire
randomized study population to avoid any possibility of selection bias.
Description of Variables
In keeping with usual practice in clinical trials, certain data were collected on the patients
at the time they entered the study. These baseline data collected on each patient included
a complete medical history, physical examination, and an evaluation of ambulatory
status. In the analysis of the clinical trial results, these data are used to assess the
comparability of treatment groups prior to receiving the treatment, that is, to assess the
adequacy of the randomization. Such baseline measurements can also be used to evaluate
treatment effects within subsets. Baseline data of greatest interest in this trial include
demographic data (patient sex and age), classification of disease into subtypes, objective
measurements of disease severity (white blood cell count, platelet count, and hemoglobin
level), and a subjective measurement of patient condition (Karnofsky scale of
performance status).
Documentation for Clinical Trial Idarubicin in AML
Page 6 of 7
Measurements were also made while each patient was being treated on the clinical trial to
determine whether a complete remission (CR) had been induced. These on study
measurements used to assign remission status included a bone marrow aspiration and
biopsy with histochemical stains to assess the continued presence of leukemic cells in the
bone marrow, a complete blood count to assess the continued presence of leukemic cells
in the circulating blood, a lumbar puncture to assess the continued presence of leukemic
cells in the brain and spinal cord, and a routine biochemical profile of the blood. Other
data were collected at specified intervals to assess patient safety and compliance.
Surviving patients were routinely followed after their last course of chemotherapy to
ascertain their continued survival and continued remission. This included patients failing
both induction courses as well as those completing all or a portion of the consolidation
period.
Determination of the primary endpoint of induction of complete remission was
conditioned on the results of the bone marrow and peripheral blood counts, and the
presence or absence of continued leukemic disease outside the bone marrow. A complete
response was defined as a bone marrow of normal cellularity or at least not hypocellular
with some evidence of normal maturation. Moreover, there had to be no more than 5%
blasts (immature blood-forming cells) in two consecutive specimens over a four week
period. Lastly, the peripheral counts had to recover and there could be no evidence of
extramedullary disease. Secondary measures of treatment effect included the number of
courses of chemotherapy required to induce remission and patient survival. Also included
in the data are measurements of the time at which a patient was judged to be in complete
remission, and indicators of whether the patient received a bone marrow transplant. Since
bone marrow transplantation might affect survival, adjustment for this last variable may
be desirable in assessing the secondary endpoint of patient survival. As the study
progressed, some patients were identified as having some form of leukemia other than
acute myelogenous leukemia, and thus they had been entered into the study by error. For
this and other reasons, some patients were judged by the primary researchers to be nonevaluable for remission status. An indicator of this evaluability is included in the data.
Table 1 displays the data available for a sample of the 130 patients in the clinical trial.
Cases with missing data for a particular variable are denoted with ‘NA’. It should be
noted that the statistical problems posed by these data revolve around the interim
analyses performed on the data as they accrued. In order to reproduce the results of those
analyses, an indicator of those cases available for analysis at the first formal analysis is
included. This analysis took place on June 30, 1988. The analysis at which the study was
terminated used data available through December 31, 1989. The patient survival data
available for an analysis must be computed from the date on study, the date of last
follow-up, the status variable, and the date of analysis. For a survival analysis, the
observation times and indicators of failure are needed. The observation time will be 0 if
the date on study is after the date of analysis. Otherwise, the observation time will be the
minimum of the difference between the date of last follow-up and the date on study and
the difference between the date of analysis and the date on study. The indicator of failure
Documentation for Clinical Trial Idarubicin in AML
Page 7 of 7
should be 1 if the date of last follow-up is prior to the date of analysis and the status
variable is 1, and it should be 0 otherwise.
Table 1: Description of variables and values for the first four cases. There are a total of 130 cases.
Missing data is denoted by `NA’.
Name
Description
First Four Cases
ptid
Patient ID
1
2
3
4
onstudy
Date on study (MMDDYY)
72384
71984
82984
90184
tx
Treatment arm (D= daunorubicin, I= idarubicin)
D
D
I
I
sex
Sex (M= male, F= female)
M
M
M
M
age
Age (years)
27
43
36
54
fab
FAB classification of AML subtype (1 - 6)
5
3
1
1
karn
Karnofsky score (0 - 100)
80
90
90
70
wbc
Baseline white blood cells (103/mm3)
179
0.9
1.8
31.9
plt
Baseline platelets (103/mm3)
51
14
71
46
hgb
Baseline hemoglobin (g/dl)
8.8
13.1
6.9
10.8
eval
Evaluable (Y= yes, N= no)
Y
Y
Y
Y
cr
Complete remission (CR) (Y= yes, N= no)
N
N
N
Y
crchemo
Courses of chemotherapy to CR
NA
NA
NA
1
crdate
Date of CR (MMDDYY)
NA
NA
NA
100884
fudate
Date of last follow-up (MMDDYY)
72984
82184
82585
10286
status
Status at last follow-up (D= dead, A= alive)
D
D
D
D
bmtx
Bone marrow transplant (Y= yes, N= no)
N
N
N
N
bmtxdate
Date of bone marrow transplant (MMDDYY)
NA
NA
NA
NA
Inclusion in June 30, 1988 analysis (Y= yes, N=
incl
Y
Y
Y
Y
no)